Improve performance of range

[eendebakpt]

Feature or enhancement

Proposal:

The range object is often used in python code. The current implementation is already efficient for integers that fit into a C long, but can be improved. The most performance critical use case is probably with a single-argument in a for loop like:

for x in range(N):
	...

For that case we can avoid several conversions of Python integers to C long and a construction of a new Python integer by adding a fast path for the single argument case (e.g. start=0 and step=1).

Alternative ideas:

• Modify the interpreter to change in range(...) in a direct construction of the iterable (skipping the range object). This would be hard to implement backwards compatible, since range could have been changed by a user to a custom implementation.
• range is read-only, so we could construct an iterable with a reference to the range object to save memory. Since the current iterable (_PyRangeIterObject) is already very small, it seems unlikely this will gain something.

Has this already been discussed elsewhere?

No response given

Links to previous discussion of this feature:

No response

Linked PRs

• gh-116477: Improve performance of range for the single argument case #116478